One of the biggest challenges to Intel’s x86 hegemony in the consumer market has been the emergence of smartphones where ARM-based processors have 90%
market share. The PC market where Intel has a stronghold is declining at (estimated) 10% annual
rate.
While a lot of coverage has been to given to Intel’s decline in the consumer market, I believe Intel has a tough road ahead in server market too. One major advantage Intel has is the amount of (legacy) code which has been written for x86 (the same thing which keeps Cobol on
life support), this advantage is diminishing rapidly due to following trends.
- The demise of gcc and rise of LLVM
For years Gcc was used, and since x86 was dominant ISA (instruction set architecture), several
optimizations were done for optimized x86 code generation.
Now, gcc is being replaced with LLVM and clang by
Google and
Apple. Given that both companies are heavily focused on their ARM-based devices, it should not come as a surprise that LLVM will have better support for ARM code generation in the long run. Rumors of Google (Intel’s 5th largest customer) working on ARM-based servers are already
out.
- VM (virtual machine) based languages A lot of server-side software development (web or otherwise) has moved to VM based languages (Java, PHP being the old leaders and Ruby, Python, Node.js the new ones).
VMs makes these platform-agnostic and hence making a move from x86 to ARM even simpler (since only the VM has to support underlying platform).
- PaaS (platform as a service) As more and more services are being moved into cloud whose underlying infrastructure is being dominated by only a few
players (Amazon, Google, etc.), it makes it, even more, easier and economical for these players to design their custom ARM processors and offer it to their customers.
